New file left out of the last commit.
[p5sagit/p5-mst-13.2.git] / lib / utf8_heavy.pl
CommitLineData
a0ed51b3 1package utf8;
cf25bb62 2use strict;
3use warnings;
a0ed51b3 4
15732964 5sub DEBUG () { 0 }
a0ed51b3 6
7sub DESTROY {}
8
5beb625e 9my %Cache;
10
86916d66 11sub croak { require Carp; Carp::croak(@_) }
12
5beb625e 13##
3a2263fe 14## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape.
15## It's a data structure that encodes a set of Unicode characters.
5beb625e 16##
17
a0ed51b3 18sub SWASHNEW {
19 my ($class, $type, $list, $minbits, $none) = @_;
20 local $^D = 0 if $^D;
15732964 21
22 print STDERR "SWASHNEW @_\n" if DEBUG;
23
cf25bb62 24 ##
25 ## Get the list of codepoints for the type.
26 ## Called from utf8.c
27 ##
28 ## Given a $type, our goal is to fill $list with the set of codepoint
5beb625e 29 ## ranges.
cf25bb62 30 ##
31 ## To make the parsing of $type clear, this code takes the a rather
2d92f8a0 32 ## unorthodox approach of last'ing out of the block once we have the
cf25bb62 33 ## info we need. Were this to be a subroutine, the 'last' would just
34 ## be a 'return'.
35 ##
5beb625e 36 my $file; ## file to load data from, and also part of the %Cache key.
37 my $ListSorted = 0;
38
cf25bb62 39 if ($type)
40 {
41 $type =~ s/^\s+//;
42 $type =~ s/\s+$//;
43
44 print "type = $type\n" if DEBUG;
45
cf25bb62 46 GETFILE:
47 {
48 ##
5beb625e 49 ## 'Is' is always optional, so if it's there, remove it.
50 ## Same with 'Category=' and 'Script='.
cf25bb62 51 ##
5beb625e 52 ## 'Block=' is replaced by 'In'.
cf25bb62 53 ##
2d92f8a0 54 my $wasIs;
55
56 ($wasIs = $type =~ s/^Is(?:\s+|[-_])?//i)
5beb625e 57 or
58 $type =~ s/^Category\s*=\s*//i
59 or
60 $type =~ s/^Script\s*=\s*//i
61 or
62 $type =~ s/^Block\s*=\s*/In/i;
cf25bb62 63
5beb625e 64 ##
65 ## See if it's in the direct mapping table.
66 ##
67 require "unicore/Exact.pl";
68 if (my $base = $utf8::Exact{$type}) {
69 $file = "unicore/lib/$base.pl";
70 last GETFILE;
cf25bb62 71 }
72
73 ##
5beb625e 74 ## If not there exactly, try the canonical form. The canonical
75 ## form is lowercased, with any separators (\s+|[-_]) removed.
cf25bb62 76 ##
5beb625e 77 my $canonical = lc $type;
78 $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g;
79 print "canonical = $canonical\n" if DEBUG;
cf25bb62 80
5beb625e 81 require "unicore/Canonical.pl";
82 if (my $base = $utf8::Canonical{$canonical}) {
83 $file = "unicore/lib/$base.pl";
84 last GETFILE;
cf25bb62 85 }
86
491fd90a 87 ##
88 ## It could be a user-defined property.
89 ##
90
09e0265a 91 my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1);
2d92f8a0 92
3a2263fe 93 if (defined $caller1 && $type =~ /^(?:\w+)$/) {
94 my $prop = $caller1 . "::" . ( $wasIs ? "Is" : "" ) . $type;
2d92f8a0 95 if (exists &{$prop}) {
96 no strict 'refs';
97
98 $list = &{$prop};
99 last GETFILE;
100 }
101 }
491fd90a 102
3a2263fe 103 ##
104 ## See if it's a user-level "To".
105 ##
106
107 my $caller0 = caller(0);
108
109 if (defined $caller0 && $type =~ /^To(?:\w+)$/) {
110 my $map = $caller0 . "::" . $type;
09e0265a 111
3a2263fe 112 if (exists &{$map}) {
113 no strict 'refs';
114
115 $list = &{$map};
116 last GETFILE;
117 }
118 }
119
cf25bb62 120 ##
3a2263fe 121 ## Last attempt -- see if it's a standard "To" name
122 ## (e.g. "ToLower") ToTitle is used by ucfirst().
123 ## The user-level way to access ToDigit() and ToFold()
124 ## is to use Unicode::UCD.
cf25bb62 125 ##
3a2263fe 126 if ($type =~ /^To(Digit|Fold|Lower|Title|Upper)$/)
cf25bb62 127 {
128 $file = "unicore/To/$1.pl";
129 ## would like to test to see if $file actually exists....
130 last GETFILE;
131 }
132
133 ##
134 ## If we reach this line, it's because we couldn't figure
135 ## out what to do with $type. Ouch.
136 ##
bc45ce41 137
138 return $type;
cf25bb62 139 }
140
491fd90a 141 if (defined $file) {
142 print "found it (file='$file')\n" if DEBUG;
143
144 ##
145 ## If we reach here, it was due to a 'last GETFILE' above
3a2263fe 146 ## (exception: user-defined properties and mappings), so we
491fd90a 147 ## have a filename, so now we load it if we haven't already.
148 ## If we have, return the cached results. The cache key is the
149 ## file to load.
150 ##
151 if ($Cache{$file} and ref($Cache{$file}) eq $class)
152 {
153 print "Returning cached '$file' for \\p{$type}\n" if DEBUG;
154 return $Cache{$class, $file};
155 }
156
157 $list = do $file;
158 }
5beb625e 159
5beb625e 160 $ListSorted = 1; ## we know that these lists are sorted
886ba366 161 }
a0ed51b3 162
15732964 163 my $extras;
485aafd9 164 my $bits = 0;
cf25bb62 165
5beb625e 166 my $ORIG = $list;
a0ed51b3 167 if ($list) {
168 my @tmp = split(/^/m, $list);
169 my %seen;
db376a24 170 no warnings;
a0ed51b3 171 $extras = join '', grep /^[^0-9a-fA-F]/, @tmp;
172 $list = join '',
df68b396 173 map { $_->[1] }
174 sort { $a->[0] <=> $b->[0] }
175 map { /^([0-9a-fA-F]+)/; [ hex($1), $_ ] }
176 grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right
a0ed51b3 177 }
178
179 if ($none) {
180 my $hextra = sprintf "%04x", $none + 1;
181 $list =~ s/\tXXXX$/\t$hextra/mg;
182 }
183
184 if ($minbits < 32) {
185 my $top = 0;
3a2263fe 186 while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) {
a0ed51b3 187 my $min = hex $1;
3a2263fe 188 my $max = defined $2 ? hex $2 : $min;
189 my $val = defined $3 ? hex $3 : 0;
a0ed51b3 190 $val += $max - $min if defined $3;
191 $top = $val if $val > $top;
192 }
193 $bits =
194 $top > 0xffff ? 32 :
195 $top > 0xff ? 16 :
196 $top > 1 ? 8 : 1
197 }
198 $bits = $minbits if $bits < $minbits;
199
200 my @extras;
201 for my $x ($extras) {
202 pos $x = 0;
1fef36c7 203 while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) {
a0ed51b3 204 my $char = $1;
205 my $name = $2;
6d6ae962 206 print STDERR "$1 => $2\n" if DEBUG;
09e0265a 207 if ($char =~ /[-+!&]/) {
a0ed51b3 208 my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really
11ef8fdd 209 my $subobj;
210 if ($c eq 'utf8') {
09e0265a 211 $subobj = utf8->SWASHNEW($t, "", 0, 0, 0);
212 }
213 elsif (exists &$name) {
214 $subobj = utf8->SWASHNEW($name, "", 0, 0, 0);
11ef8fdd 215 }
216 elsif ($c =~ /^([0-9a-fA-F]+)/) {
217 $subobj = utf8->SWASHNEW("", $c, 0, 0, 0);
218 }
bc45ce41 219 return $subobj unless ref $subobj;
a0ed51b3 220 push @extras, $name => $subobj;
221 $bits = $subobj->{BITS} if $bits < $subobj->{BITS};
222 }
223 }
224 }
225
15732964 226 print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG;
a0ed51b3 227
5beb625e 228 my $SWASH = bless {
a0ed51b3 229 TYPE => $type,
230 BITS => $bits,
231 EXTRAS => $extras,
232 LIST => $list,
233 NONE => $none,
234 @extras,
235 } => $class;
5beb625e 236
237 if ($file) {
238 $Cache{$class, $file} = $SWASH;
239 }
240
241 return $SWASH;
a0ed51b3 242}
243
244# NOTE: utf8.c:swash_init() assumes entries are never modified once generated.
245
246sub SWASHGET {
035d37be 247 # See utf8.c:Perl_swash_fetch for problems with this interface.
a0ed51b3 248 my ($self, $start, $len) = @_;
249 local $^D = 0 if $^D;
a0ed51b3 250 my $type = $self->{TYPE};
251 my $bits = $self->{BITS};
252 my $none = $self->{NONE};
15732964 253 print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG;
a0ed51b3 254 my $end = $start + $len;
255 my $swatch = "";
256 my $key;
257 vec($swatch, $len - 1, $bits) = 0; # Extend to correct length.
258 if ($none) {
259 for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none }
260 }
261
262 for ($self->{LIST}) {
263 pos $_ = 0;
264 if ($bits > 1) {
265 LINE:
3a2263fe 266 while (/^([0-9a-fA-F]+)(?:[ \t]([0-9a-fA-F]+)?)?(?:[ \t]([0-9a-fA-F]+))?/mg) {
267 chomp;
268 my ($a, $b, $c) = ($1, $2, $3);
269 croak "$type: illegal mapping '$_'"
270 if $type =~ /^To/ &&
271 !(defined $a && defined $c);
272 my $min = hex $a;
273 my $max = defined $b ? hex $b : $min;
274 my $val = defined $c ? hex $c : 0;
a0ed51b3 275 next if $max < $start;
6d6ae962 276 print "$min $max $val\n" if DEBUG;
a0ed51b3 277 if ($none) {
278 if ($min < $start) {
710250cb 279 $val += $start - $min if $val < $none;
a0ed51b3 280 $min = $start;
281 }
282 for ($key = $min; $key <= $max; $key++) {
283 last LINE if $key >= $end;
6d6ae962 284 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3 285 vec($swatch, $key - $start, $bits) = $val;
286 ++$val if $val < $none;
287 }
288 }
289 else {
290 if ($min < $start) {
291 $val += $start - $min;
292 $min = $start;
293 }
294 for ($key = $min; $key <= $max; $key++, $val++) {
295 last LINE if $key >= $end;
6d6ae962 296 print STDERR "$key => $val\n" if DEBUG;
a0ed51b3 297 vec($swatch, $key - $start, $bits) = $val;
298 }
299 }
300 }
301 }
302 else {
303 LINE:
99a6b1f0 304 while (/^([0-9a-fA-F]+)(?:[ \t]+([0-9a-fA-F]+))?/mg) {
3a2263fe 305 chomp;
a0ed51b3 306 my $min = hex $1;
3a2263fe 307 my $max = defined $2 ? hex $2 : $min;
a0ed51b3 308 next if $max < $start;
309 if ($min < $start) {
310 $min = $start;
311 }
312 for ($key = $min; $key <= $max; $key++) {
313 last LINE if $key >= $end;
6d6ae962 314 print STDERR "$key => 1\n" if DEBUG;
a0ed51b3 315 vec($swatch, $key - $start, 1) = 1;
316 }
317 }
318 }
319 }
320 for my $x ($self->{EXTRAS}) {
321 pos $x = 0;
09e0265a 322 while ($x =~ /^([-+!&])(.*)/mg) {
a0ed51b3 323 my $char = $1;
324 my $name = $2;
15732964 325 print STDERR "INDIRECT $1 $2\n" if DEBUG;
1fef36c7 326 my $otherbits = $self->{$name}->{BITS};
327 croak("SWASHGET size mismatch") if $bits < $otherbits;
328 my $other = $self->{$name}->SWASHGET($start, $len);
329 if ($char eq '+') {
330 if ($bits == 1 and $otherbits == 1) {
331 $swatch |= $other;
a0ed51b3 332 }
1fef36c7 333 else {
334 for ($key = 0; $key < $len; $key++) {
335 vec($swatch, $key, $bits) = vec($other, $key, $otherbits);
a0ed51b3 336 }
1fef36c7 337 }
338 }
339 elsif ($char eq '!') {
340 if ($bits == 1 and $otherbits == 1) {
341 $swatch |= ~$other;
342 }
343 else {
344 for ($key = 0; $key < $len; $key++) {
345 if (!vec($other, $key, $otherbits)) {
346 vec($swatch, $key, $bits) = 1;
a0ed51b3 347 }
348 }
349 }
1fef36c7 350 }
351 elsif ($char eq '-') {
352 if ($bits == 1 and $otherbits == 1) {
353 $swatch &= ~$other;
354 }
355 else {
356 for ($key = 0; $key < $len; $key++) {
357 if (vec($other, $key, $otherbits)) {
358 vec($swatch, $key, $bits) = 0;
a0ed51b3 359 }
360 }
361 }
362 }
09e0265a 363 elsif ($char eq '&') {
364 if ($bits == 1 and $otherbits == 1) {
365 $swatch &= $other;
366 }
367 else {
368 for ($key = 0; $key < $len; $key++) {
369 if (!vec($other, $key, $otherbits)) {
370 vec($swatch, $key, $bits) = 0;
371 }
372 }
373 }
374 }
a0ed51b3 375 }
376 }
15732964 377 if (DEBUG) {
a0ed51b3 378 print STDERR "CELLS ";
379 for ($key = 0; $key < $len; $key++) {
380 print STDERR vec($swatch, $key, $bits), " ";
381 }
382 print STDERR "\n";
383 }
384 $swatch;
385}
386
3871;